A Wireless Sensor Network Platform

Optimized for Assisted Sustainable Agriculture

Abel Rodriguez de la Concepcion, Riccardo Stefanelli, Daniele Trinchero
iXem Labs Foundation - DET
Politecnico di Torino
Torino, Italy
info@iXem.polito.it
Abstract The paper illustrates an efficient wireless sensor
network platform, suitable for application to assisted agriculture
in (but not only in) developing Countries and remote regions.
The platform has been conceived in order to minimize power
consumption, during all the phases of data acquisition, sampling,
and compression, with an efficient and performing
communication protocol, with extended transmission range and
radio coverage optimization. Sensor nodes have been provided
with energy harvesting facilities, to avoid any need for direct
power supply or battery replacement. The resulting nodes are
consequently autonomous, easy to locate and relocate, and
scalable. A further work has been done to minimize dimensions
and costs, in order to deploy capillary installations. Furthermore,
thanks to the work done from the side of the channel
optimization, it has been possible to acquire not only standard
environmental parameters, but also high definition pictures.
Images of plants, trees, as well as fruits and leaves are taken
every hour, and forwarded to a central gateway, interfaced with
the Internet. A team of agronomists and biologists checks the
state of the cultivation from remote, providing the farmer with
continuous assistance at a reasonable cost. This is extremely
important in Developing Countries, taking into account that in
those locations experts cannot reach the fields and cannot
provide the farmer with specialized, continuous consultancy,
both for economical and logistic reasons. In a global scenario,
where new diseases arise rapidly and continuously, the remote
assistance provided by an expert can minimize farmers risks to
loose his harvest and reduce his revenues. The set of
environmental parameters, together with the visual collection of
cultivation conditions, is useful also to generate a culture
database, particularly useful in developing regions, where there
is almost never historical recorded trace, in particular about
possible associated diseases and infections. Last but not least, the
platform allows a significant improvement of the sustainability.
Thanks to the assistance of the agronomist, the farmer can
minimize the use of pesticides and chemicals, as well as reducing
the number of additional treatments, resulting in significant
advantages, in terms of production costs and organic quality.

competence [1]. In the recent years, because of weather

changes and vehicular transportation of infections, the
historical know how handed down by the previous generations
has revealed to be insufficient, and new diseases and infections
arise almost every season. Adopting external consultancy by an
expert is not possible, for economical reasons, but also for
logistic complexity: regions lack of experts and the few
available cannot travel regularly to any field. If the farmer has
the possibility, he may adopt chemical treatments, which are
very expensive and reduce his net margin. In this way, the
sustainability of the cultivation, as well as the natural quality of
the goods, is drastically affected.

Keywordscomponent; Internet of Things, Wireless Sensor

Networks, Intelligent Agriculture

Even more, environmental sustainability is becoming a

relevant objective in any agriculture cultivation [1].
Sustainability has several facets, the most relevant one being
represented by the effort to increasingly contain the use of
chemical agents, in particular pesticides. I.e., the European
Union is promoting, any agriculture practice that eludes the use
of pesticides. REACH (Registration, Evaluation, Authorization
and Restriction of Chemical substances) is a European
Community regulation to improve the protection of human

I.

It is a matter of fact that the only possibility to seriously

help the farmer is represented by an autonomous, independent,
tentatively continuous analysis of the cultivation state, with
regular feedbacks about the actions that are needed. Reporting
the acquired data over the Internet, could allow a troupe of
agronomists and biologists to assist the farmer, or a large
number of farmers, in real time, for a huge number of
cultivations, cancelling logistics and organization costs: the
consultancy cost would be shared among many farmers, and no
transportation (or very limited one) would be required.
Both in organic and conventional agriculture. the most
relevant instruments to operate a pervasive control, is
represented by a capillary, continuous and constant monitoring
of any event occurring to the cultivation, both from the side of
the plants and the one of the surrounding environment. In
developed Countries, this is possible by activating periodic in
field controls. In Developing Countries, this would generate
relevant expenditures, (personnel, equipment, fuel) and
logistics complexity (travelling on dirty roads, bad weather
limitations, personnel availability). All these factors limit the
frequency, and consequently the efficiency, of the control,
which can be performed rarely during the season, while
infection conditions may vary rapidly during the day.

INTRODUCTION

In a Developing Country, the farmer, who is in the same

time the plant cultivator, the good stocker and the product
seller, is the only responsible for the cultivation, where he
mainly applies his empirical but effective agronomic

978-1-4799-7193-0/14/$31.00 2014 IEEE

159

IEEE 2014 Global Humanitarian Technology Conference

health and the environment [3], which calls for the substitution
of dangerous chemicals with safer alternatives. In September
2008, the European Union issued new and revised Maximum
Residue Limits (MRLs) for the roughly 1,100 pesticides ever
used in the world [4].
For all the reported reasons, we have addressed the problem
by constructing a wireless sensor network (WSN) platform,
which reports data over the Internet.
II.

WSN PLATFORMS FOR SUSTAINABLE AGRICULTURE

Lets consider viticulture as an example. From the

agronomic point of view, viticulture uses an already well
established method based on pheromone observation on
lobesia botrana, and sticky traps for scaphoideus titanus.
Unfortunately, for the most common fungal diseases, there are
no available methods other than the constant monitoring by a
qualified technician. The monitoring can be efficiently piloted,
by knowing in advance the growing conditions, in particular
the temperature, the humidity, and leaf wetness.
For this reason, during the last ten years, several attempts
have been made, to set up meteorological stations within
vineyards [5], In the usual application, stations are as frequent
as the number of Municipalities involved, and each station may
control, using a wireless channel, a number of sub-stations,
which detect information about leaf wetness, with at least one
substation every 25 hectares [6].
WSNs have significantly contributed to a better and
capillary monitoring of the vineyards/ In general, WSNs are
used to sense (control) parameters of various kinds, in very
different application contexts: industry, services, trade, urban
activities, logistics, and of course agricultural production. Both
in the market and in the scientific literature, various examples
of systems and platforms exploiting wireless communication
schemes can be found. Some of them make use of international
standards like Zigbee, Ultrawideband, Bluetooth, specifically
dedicated to the acquisition of relevant quantities of data,
respectively in industrial environments, in electromagnetically
noisy environments, and in relatively small scenarios. Other
(AdHoc, MANET) are constructed following the specific needs
of the implementer, and they are usually based on the need to
minimize energy consumption, with typically very small data
to be transmitted. All mentioned systems make us of
unlicensed frequency bands, to minimize maintenance costs
and management.
With the advent of WSNs, it has been possible to increase
the number of sensing nodes [7], transforming the traditional
meteorological system into a more pervasive control of the
cultivation [8]. The required computing effort is relatively
small, as the acquisition frequency should be as frequent as one
per hour [9].
More recently [10], WSNs have been proposed for an
enhanced vineyard monitoring, which complements
meteorological and environmental information with image
acquisition. Plant images are acquired in strategic points
chosen by the agronomist. In this way, the agronomic
observation becomes almost continuous and the agronomist
analysis can be automatically transferred inside the node, by

Fig. 1. Example of application of the proposed system

implementation of image processing techniques. The

inconvenient of the realization proposed in [10] is represented
by the use of Wi-Fi, which allows the transportation of large
amount of data, but unfortunately generates too much power
consumption.
III.

OUR WSN ARCHITECTURE

Our proposal is an ad-hoc platform similar to the ones used

to acquire environmental and meteorological data, where the
manufacturing cost of the sensor node is minimized, by an
efficient selection of the system components. The aim is the
realization of an ergonomic, low-cost, miniaturized platform,
characterized by low environmental impact, low power
consumption and high performance. Clearly, this monitoring
solution cannot prevent all possible diseases, and the efficiency
of a consultant through the Internet is much lower, compared to
the one that he might give on site. Nevertheless, it appears as
an effective trigger to connect farmers and experts, helping the
formers in their daily operations.
Moreover, on the same network, in relevant places selected
by the agronomist, high definition cameras are installed, in
order to provide the expert with adirect photographic
information. The scope is the one described in [10], without
using wide band transmitting systems. In a way, by adding the
image acquisition facility to an ad-hoc wireless sensor nodes
terminal, we redefine the role of an ad-hoc WSN, overcoming
the typical acquisition and transmission schemes that allow
temperature, humidity, wetness monitoring.
As a consequence, the challenge is represented by the need
to set up an ad hoc system, to be installed within leaves, an
environment that is unfavorable to the radiofrequency
propagation, with a capability to transfer data at a transmission
rate typical of the wideband data access systems. The
acquisition must be optimized, and the best compromise
between the needs of the user and the constraint to limit energy
consumption and space occupation must be reached: the only
chance to favor an application on a large scale.
The technical concept is illustrated in Fig. 1, where we
recognize a typical ad-hoc wireless platform, where sensor

Fig. 2. Sensor Node Scheme

nodes are installed among the plants, and can be used not only
to acquire, but also to establish a multi-hop communication
structure, until the information is reported to the base stations,
and from there to the Internet.
IV.

Fig, 4. Base Station Implementation

BASE STATON AND REPEATERS

The base station (BS) and the repeaters have been

implemented in a modular way, by minimizing the equipment
cost, optimizing flexibility and scalability, and aiming to
reduce implementation times. A single board computer (SBC)
running an open source operating system has been used as
main data processing unit. The communication component has
been managed using a transceiver manufactured by Texas
Instruments, model CC1110. As an SBC, a Raspberry model B
has been implemented. The BS bloc diagram is reported in
Fig.3. Fig. 4 shows the implemented device, where the radio
interface has been interfaced to the SBC via a standard serial
port. The transceiver has the characteristics listed in Tab. I,
and it has been configured to work, alternatively, at 865 MHz
or 433 MHz, depending on the system requirements and the
propagation scenario. The higher frequency is preferable when
sensors are located in line of sight, while the lower one
simplifies the installation process in a hilly, complex scenario.

TABLE I.

TRANSCEIVER DATA

PARAMETER
Frequency bands [MHz]
TX Current consumption [mA]
RX Current consumption [mA]
Sleep power consumption [uA]
Data rate [kbps]
Sensitivity [dBm]
Output power [dBm]
Frequency modulation
Amplitude modulation

V.

VALUE
315/433/868/915
31
22
0.6
1.2 - 250
-110 (@ 2.5 Kbps)
10
2-FSK/GFSK
OOK/ASK

SENSOR NODES

The sensor node (SN) is the relevant component of the

system. The node controls the environmental sensors and the
camera (when present). The data acquisition is managed by the
transceiver, together with the transmission to the BS. The
antenna is specifically designed and integrated on board, and
the power supply is provided by a dedicated photovoltaic
harvester. The sensor scheme is reported in Fig. 2.
A. Environmental Sensor
Several sensors can be managed by the transceiver. For our
first implementation, we have selected localized temperature
and humidity units. The chosen unit is produced by Sensirion,
Model SHT21 (Fig. 5). Tab II lists the main sensor
characteristics.
TABLE II.

Fig. 3. Base Station Scheme

PARAMETER
Data interface
Power consumption
Humidity range
Temperature range
Footprint

ENVIRONMENTAL SENSOR DATA

VALUE
I2C
2.3uW
0 - 100% RH
-40 - +125 C
3 x 3 mm

Fig. 5. Example of application of the proposed system

B. Cameras
The camera used is the OV5642 from OmniVision. It is
able to capture and process pictures with different resolutions,
ranging from 320 X 240 to 2592 X 1944 pixels. Furthermore, it
is able to compress images and to generate a JPEG data stream.
An external microcontroller can be interfaced in order to set the
camera. The image acquisition processing is carried out
without external intervention and the image data are sent to the

Fig, 7. The photovoltaic harvester

microcontroller through a serial interface. The module has a

power save mode which aims to increase the energy autonomy,
ranging from 270mA in active mode (when the image is taken)
to 25uA in power save mode.
C. Transceivers
As a transceiver, the same component exploited for the BS
has been implemented. Characteristics are the same listed in
Tab I, even if in this case the transceiver is used to
acquire/store data from the sensors and the camera, in
addiction to commands reception and data transmission.
D. Antenna
The typology chosen for the mentioned application is the
resonating magnetic radiator, which is significantly efficient in
lossy media, where the field region is characterized by a wave
impedance as small as the one of the medium surrounding the
antenna. The radiator is omnidirectional, and consequently the
communication range is limited. In RFID and WSNs
technology, this limitation does not affect the performance of
the system, as those systems are in any case characterized by
EIRP limitations.

Fig. 6. Antennas working at 433 MHz (top) and 868 MHz (bottom)

To minimize the circuit dimension, we have identified an

efficient magnetic antenna that allows the realization of
matched loop radiators, by embedding the matching circuit
inside the loop [11]. An example is shown in Fig.6, bottom left
(868 MHz), where the antenna is a standard circular loop and
the two curvilinear transmission lines are inserted in series, to
cancel the imaginary part of the impedance that can be
measured at the loop terminals. To reduce the dimensions of
the antenna, allowing an extended minimization of space
occupation, more complex geometries have been exploited. In
this way, it has been possible to synthesize small antennas,
embedded inside circumferences having a radius down to 60%
of a standard circular loop. To this purpose, a geometrical
shape similar to a flower, where the number of petals is
optimized to increase the current path, has been introduced, as

Fig. 8. The SN nude, with embedded radiator (left) and external

radiator (right)

shown in Fig 6, top (433 MHz) and bottom right (868 MHz).
As one can see, the 60 % size reduction is almost confirmed.
E. Harvester
According to the agronomist constraints, one image per
hour is a good starting point to monitor the condition of the
plant. Consequently, the duration of the full transmission of the
image depends on the resolution of the camera and the data
rate. As an example, for the calculation of the energy budget
we might consider a duration of about 6 minutes for a high
resolution image transmitted in average coverage conditions.
For the following 54 minutes the system sleeps, with a power
consumption as small as 0.8uA. Consequently, the total power
consumption is 27.5 mW/h. Assuming that the system works
10 hours per day, the total power consumption is 330 Mw/day.
In order to minimize the costs, a rechargeable battery with a
capacity equal to 850 mA/h @ 3.7 V has been used. This
battery is able to power the system for about 8 days. So, a solar
panel has been chosen as energy harvester in order to recharge
the battery. According to this power budget a panel with
dimensions 7 x 5.5 cm has been chosen. Assuming 3 sun hours
per day, the panel provides 500 mW/day. In this configuration
the system is complete and fully autonomous without time
limitations: the batteries are almost perpetual and they can be
recharged regularly by the solar panel.

Fig. 10. A first example of image acquired by a SN in Sorii San Lorenzo,

Barbaresco, Italy

F. Full Sensor Node

The Sn is assembled by packing all the mentioned devices
together, taking care of minimizing the overall dimensions.
The package has been chosen in order to survive to normal
working operations within the vineyard, together with the
capability to resist to any kind of weather condition. Fig. 8
shows an example of the SN, nude. Fig 9 shows the same,
within its own package, which has been designed in order to be
waterproof.
VI.

VII.

Fig. 9. Two SNs installed in a vineyard. Left: with internal harvester,

right, with external harvester

INTERNET ACCESS

Data, once acquired, are transferred to the BS, and from

there they are copied on a remote database, via an ftp channel
over the Internet. The communication channel is realized by
means of a standard 3G modem, directly implemented on the
BS SBC. A web interface has been implemented, in order to
allow the agronomist to access the database, to review images,
to select the relevant ones, comparing them in a dedicated
framework. Moreover, the user has the possibility to analyze
the evolution of the environmental parameters, generating
automatic graphs and cross-comparing the evolution of the
vineyard along the years (in the future, when enough data will
be available).
THE CURRENT EXPERIMENT

The first experiment has been implemented in the

Municipality of Barbaresco, Piedmont, in the Sorii San
Lorenzo vineyard, which belongs to the wine producer Gaja.
The same experiment could be replicated anywhere else in the
World. One BS has been implemented in the farmhouse, while
sensors have been deployed up the hill. The vineyard is
represented in Fig. 1, showing a picture taken from the
farmhouse, exactly from the same position where the BS has
been installed. SNs have been deployed in several positions,
both in line of sight and not. For this reason, the
communication between the BS and the SNs has been
implemented in an adaptive way, in order to optimize the living

this way, we have demonstrated that the remote monitoring has

given the possibility to cancel the chemical treatments, and
reduce for a 50% the contact treatments. Even if these results
should be validated on a larger scale and the system should be
applied to a real Developing scenario, the experiment shows
the applicability of the platform and its economical
sustainability. In terms of man power, we have estimated the
time required by the expert to analyze data coming from one
hectare: it corresponds (in average) to 1 minute per day. In
terms of costs per season, it corresponds to 3 hours per hectare
per season, an amount relevantly less than the one
corresponding to the saved treatments. It is then clear how the
WSN application represents not only an advantage for the
sustainability, but also a relevant money saving for the farmer.
IX.
Fig. 11. A second example of image acquired by a SN in Sorii San
Lorenzo, Barbaresco, Italy

time of each SN: every SN transmits using the highest

modulation rate permitted by its distance/visibility from the
BS. In this way, 2 MB pictures are transferred in a time
comprised between 45 and 745 seconds. To optimize the
transmission time, we are now implementing a multi-hop
system making use of repeaters, which will allow a line of
sight condition to all SNs.
Fig. 10 and Fig. 11 show examples of data acquired using
the monitoring system. A complete gallery of the images
collected during the experimentation is available for
consultation on the iXem Labs website [12]. The experiment
has been activated on May 1st, 2013, and since then, for the
whole season (until November 30th), the SNs have been active
and in good working conditions. The energy consumption has
been kept under constant control, and the very small harvester
has been able to supply power to avoid any loss of data. From
the side of battery duration, the experiment cannot be
considered significant, as its duration was limited to one
season. Nevertheless, the same harvesting configuration was
previously adopted on a WSN platform used to monitor
avalanches [13], where a duration of at least five years was
testified, even if weather conditions where extremely tough and
much worse than the ones found in a farm all over the World.

The work presented in this paper proposes a remote

monitoring solution to provide farmers in Developing
Countries with high level, low cost assistance to accelerate the
efficiency of their cultivations, improving the environmental
sustainability of the whole process and the quality of the
production. The validation, for the moment, took place in Italy,
where all parameters typical of a Developing scenario were
reconstructed. The technological platform, the agronomical
result, and the economical outcome were evaluated and
validated on a lower scale. The next step will involve a large
scale set-up in a real Developing scenario.
ACKNOWLEDGMENT
This work has been possible, thanks to the collaboration of
Cantine Gaja, who provided access and use to their vineyards,
and Stazione Sperimentale per la Viticoltura Sostenibile, who
provided suggestions and agronomical consultancy. The
authors thank Dr. Angelo Gaja, Mr. Giorgio Culasso, and Dr.
Ruggero Mazzilli for their relevant and fundamental support.
REFERENCES
[1]
[2]

[3]

VIII. ECONOMICAL EVALUATION

With reference to the specific experiment that we have run
during 2013, we have identified the minimum number of
sensors needed for an adequate control of the cultivation: 2 Sn
per hectar. We have estimated a fabrication cost of about 65
USD per SN, with one BS needed every 20 SN (cost per BS 78
USD). The average cost for a chemical (systemic) treatment is
at least 120 USD per hectar, with about 10 treatments required
per season. The average cost for a non-chemical (contact)
treatment is at least 50 USD, with about 15 treatments required
per season. During season 2013, in our experiment in
Barbaresco, we have compared to similar productions, the
former provided with remote monitoring, the latter without. In

CONCLUSIONS

[4]
[5]

[6]
[7]

[8]

CRUMP, Larry; MASWOOD, S. Javed (ed.). Developing countries and

global trade negotiations. Routledge, 2007.
Gold, Mary. 1999. Sustainable Agriculture: Definitions and Terms.
Special Reference Briefs Series no. SRB 99-02 Updates SRB 94-5
September 1999. National Agricultural Library, Agricultural Research
Service, U.S. Department of Agriculture.
"What is REACH?". European Commission. 21 May 2010. Retrieved 29
July 2010.
European Commission (2008)Plant Protection - Pesticide Residues Fact
Sheet,
Raynal, M., S. Guittard, C. Debord, A. Davy, T. Plancq, M. Vergnes, S.
Codis, and O. Hebrard. "Decision Support Systems in viticulture:
Current situation and perspectives."
Langellier, F., and L. Panigai. "A climatic synthesis: meteo network in
the Champagne vineyard (France)." Vigneron Champenois (1999).
Suri, Ankur, S. Sitharama Iyengar, and Eungchun Cho. "Ecoinformatics
using wireless sensor networks: An overview." Ecological Informatics 1,
no. 3 (2006): 287-293.
Matese, A., Di Gennaro, S. F., Zaldei, A., Genesio, L., & Vaccari, F. P.
(2009). A wireless sensor network for precision viticulture: The NAV
system. Computers and Electronics in Agriculture, 69(1), 51-58.

[9]

Burrell, Jenna, Tim Brooke, and Richard Beckwith. "Vineyard

computing: Sensor networks in agricultural production." Pervasive
Computing, IEEE 3, no. 1 (2004): 38-45
[10] Lloret, Jaime, Ignacio Bosch, Sandra Sendra, and Arturo Serrano. "A
wireless sensor network for vineyard monitoring that uses image
processing." Sensors 11, no. 6 (2011): 6165-6196.

[11] Carvallo N, Stefanelli R., Trinchero D., Wireless interfaces for sensor
networks embedded in tough environments, URSI EMTS, Hiroshima,
Japan, May 20-25, 2013.
[12] www.iXem.polito.it
[13] CARTA, Alessandra, et al. Radiofrequency sensors for snow conditions
monitoring and real time avalanche alerts. In: Wireless Sensors and
Sensor Networks (WiSNet), 2013 IEEE Topical Conference on. IEEE,
2013. p. 49-51.